Booster Cogging: Synchronization with the Main Injector for NuMI & Slip-Stacking Robert Zwaska University of Texas at Austin Budker Seminar July 27, 2004

Booster Cogging:Synchronization with the Main Injector

for NuMI & Slip-Stacking

Robert Zwaska

University of Texas at Austin

Budker SeminarJuly 27, 2004

July 27, 2004 Robert ZwaskaBooster Cogging

2

IntroductionMyself

• Graduate Student at the University of Texas Sine 1999

• NuMI/MINOS Since 2000

• Booster Group Since 2003

• Fermi Ph.D. program Since 2003 or 2004

People Involved• Sacha Kopp

Advisor in Texas

• Eric Prebys Supervisor at FNAL

• Bill Pellico Cogger at FNAL

• Bob Webber Earlier Cogger

• Rich Meadowcraft, Todd Sullivan, Andrew Feld, Jim Lackey, Alberto Marchionni, Alex Waller, Craig Drennan Have helped out


3

OutlineI. Motivation

i. Multi-Batch Operation

ii. Booster Losses

II. Notching the Booster BeamI. Extracting with the Notch

III. Sources of Slippage

IV. Cogging Method

V. Examples of Issues Encountered


4

The Fermilab Booster• Multiple-turn injection of H- ions

Carbon stripping foil Typically 12-14 turns

• Accelerates 400 MeV 8 GeV• Fast cycling synchrotron

Resonant magnet ramp Frequency of 15 Hz locked with wall-

socket• Circumference of 474 m• Beam bunched at 37 MHz

Harmonic number = 84 53 MHz at extraction

• 18 RF Stations 0.9 MV• Accelerates in 33 ms

T = 5.45• 5-6 x 1012 protons/pulse

80 – 90 % efficient• Single turn extraction

Two extraction regions

Injection

Extraction to Main Injector /

MiniBooNE

Extraction to Beam Dump

The Main Injector• Accepts 8 GeV protons from the Booster• Accelerates to 120 GeV

Uses 53 MHz from the Booster 4 MV of RF power

• Circumference of 3319 M 7 x the length of the Booster


6

NuMI/MINOS

• Long-Baseline experiment Neutrino oscillations measurements 735 km

• Uses 120 GeV protons from the Main Injector

• Designed for 400 kW ave. power Will be closer to 250 kW as start

• Seriously statistics limited Initially needs 8 x 1020 protons

• Will start December 2004

Neutrinos at the Main InjectorMain Injector Neutrino Oscillation SearchN

eutrino beam


7

Slip-Stacking• Combines two Booster

batches longitudinally In the Main Injector

• To be used, initially, for antiproton production Part of Run II Upgrades

Tim

e

Longitudinal Position


8

Multi-Batch Operation

Batch 1 (PBar)

Batch 2

Batch 3

Batch 4

Batch 5

Batch 6

Booster

Main Injector

½ Batch(empty) ½ Batch

(empty)

• Two beams must be accelerated together in the Main Injector Extracted to PBar & NuMI

• Requires “multi-batch” operation 6 batches from the Booster

NuMI


9

Radiation Issues

• Radiation is the operational limit on Booster operation Limits integrated number of protons

• Residual activation in the tunnel Radioisotopes created by showers Long lived isotopes limit how much

maintenance can be done in the tunnel

• Damage of beam components• Prompt radiation from the

showering of lost protons Radiation scales with energy and

number of protons lost Very small amount penetrates the

shielding


10

Extraction Need for a Notch• Extraction kicker magnet has a

risetime of ~ 40 ns Only ~ 10 ns between bunches

• Beam lost at 8 GeV Lost on septum magnet at extraction Intolerable for extended operation

• Instead, beam is removed at 400 MeV Create a “notch” in the beam Activation scales, roughly, with beam

energy lost• Loss reduced by 95% (from this

source) Position of losses can be chosen as a

non-critical area• Extraction kicker firing must be

synchronized with the notch Easy with one batch…

84 RF bucketsaround circumference

Notch

Booster


11

Notching• Induce a large vertical betatron

oscillation• Uses a kicker magnet• Displacement goes as:

y ~ 90o / period• Most of the beam is deposited in

the third magnet• Kick must be large enough to

scrape off beam Beam is stiffer at higher energy

Notcher

Pinger

)sin( Vy

py


12

Notching for Multibatch- Cogging -

84 RF bucketsaround circumference

NotchBooster

Main Injector

Previous injectedPrevious injectedBooster batchBooster batch

• Booster extraction to MI must be synchronized with the notch• Extraction must also be synchronized to the beam already

circulating in the Main Injector• Problem: The Booster and Main Injector are not necessarily

synchronized Booster beam “slips” relative to the Main Injector Amount of slippage is not consistent cycle-to-cycle

• “Cogging” is forced synchronization of beams No Booster flattop available to fix at the end Active feedback is necessary during acceleration


13

Raw position

Following the Notch

• RF buckets slip at a rate fMI – fB

Initially 15 MHz Notch wraps around the Booster

many times

• Extraction with one batch Count RF buckets to make a marker Extract on marker

• Tunable delay

Reset Main Injector

• With several batches Possible if total slippage is exactly

the same Requires 1 in 100,000 consistency

6/7 resonance


14

Measuring Slippage• Monitor Notch position throughout

the cycle• Use Main Injector RF as a standard

clock• Booster RF frequency varies with

energy 38 53 MHz

• Start counting on Main Injector revolution marker

• Stop Counting on Booster revolution marker

• Makes a table of positions (tripplan)

588 buckets @ 52.8114 MHzMain Injector

Revolution Marker

84 buckets @38 – 53 MHz

Booster Revolution Marker (Notch)

Booster RF 38 – 53 MHz

in 1in 2in


15

Raw position Relative to baseline

~ 3 turns

Relative Slippage• Slippage in a cycle varies by > 200 buckets

About 3 circumferences• Notch is essentially at a random position w.r.t the beam

circulating in the Main Injector


16

Sources of Slippage

• Any change in the difference of frequencies (fMI, fB) will cause slippage

• Some errors in the Booster can be parameterized as a change in p(t)

• Slip rate: (buckets/ time) 15 MHZ 0

• (inj. ext.)

• Variations in wall socket frequency has long been suspected as a source of error Booster 15 Hz is line f ÷ 4

2

2

2

)(MI

MIBMI1

11

p

mcfff

t

tdttS0

'' )()(

)2cos()( 10 ftpptp

GeV/c 9223.4)( ie21

0 ppp

GeV/c 9666.3)( ie21

1 ppp


17

Error Characterization• Any perturbation to (t) will resulting

slippage• Some errors can be parameterized as a

change in p(t) Each has a particularly shaped S(t) curve

MaximumMagnet Current

Magnet Frequency

Timing

MinimumMagnet Current

Time (ms)

S(t

) (

norm

aliz

ed)

t

tdx

ddtxtS

0

'' )()(

• Several possible errors shown below: Timing: 1 s 15 bucket slip Magnet Frequency: 1 mHz 6 bucket slip Minimum Magnet current (pi): 1/10,000

10 bucket slip Maximum Magnet current (pe): 1/10,000

7 bucket slip


18

sample

Notchradial feedback

Cogging Method• Use the early part of the cycle to

predict net slippage before extraction Place the notch anticipating further

slippage (few ms into cycle)

• After transition slippage can be induced by changing the radial position of the beam Changes the feedback of the Low

Level RF systems Changes the circumference, and thus

period, of a revolution Active feedback can correct to zero

error

r

~ 3 turns


19

Intelligent Notching• Depends on consistent slippages signature

i.e. source of slippage

• Application of the notch 5 ms later can reduce range

~ 3 turns

~ 20 buckets


20

First Tests of Cogging• Notch using prediction• Radial Feedback late in the cycle• r = k ∙ S

Exponential damping k 0.2 mm / bucket e-folding time 10 ms

NotchRadial

Feedback

Notch

RadialFeedback

Doesn’t get to zero

Recent Cogging in the Booster• Shown are the second batches of a slip-stacking cycle

All slippage is to one side Radial feedback is significant

• Radial offset goes to a constant value to get closer to zero 90% with 0 error Other 10% with ± 1 (out of ~ 70)

Notch

RadialFeedback 0 ms 33 ms

0e12-5 mm

2.4e12+5 mm

15 ms 35 ms

Intensity

Radial PositionsOf multiple cycles


22

Beam in the Main Injector

• These are Booster Batches lined up in the Main Injector Two different beam tests

• Kicker extraction is synchronized with beam in Main Injector Not to the notch in Booster

• Upper set is uncogged Notch appears randomly within

the batches

• Second batch is cogged Notch appears at a consistent

position with each batch Only the timing of the notcher

need be adjusted• Wouldn’t be visible otherwise

Tim

e

Tim

e

Notches


24

Ping Notching

• Current notch knocks out the beam with one kick Works less well at slightly higher momenta Activates one region with the loss

• This loss is irreducible and scales with the number of protons

• Alternative: Apply a series of smaller kicks (pings) Similar, in concept, to anti-damping

• But, no feedback Takes a few hundred turns Uses a solid-state driver Losses go into collimators

• Need to account for tune spreads and shifts


25

Broadband Pinging Simulation

Applying a broadband sequence of pings allows a

range of tunes to be kicked out


26

Intensity in the Main Injector

• Since commissioning, MI has run with only 5 x 1012 protons in it 1/7 of the ring

• With NuMI, there will be at least 3 x 1013

5+1 x 1012

Booster per-batch intensity may be higher Stacking schemes might allow another 50%

• Numerous problems need to be addressed Foremost are instabilities due to beam-loading and other collective effects

• This is just to get to our planned intensities, higher will require significant upgrades


27

Time

Phase

A View of Slip-Stacking


28

GMPS• Booster magnets are part of a resonant circuit

DC & AC componentsAC is part of a LRC circuit with choke & caps in

tunnel

• Circuit controlled by “GMPS” (Gim-pis)Gradient Magnet Power Supply


29

Settings

Read-backs

Feedback


30

GMPS Regulation• Observed Oscillation of minimum

current Occurs over a train of cycles

• Suggestive of line draw, and compensation Feedback with t 400 ms

• Hypthesized cause: Pulsed devices drag line current down Lower line current reduces power

input to GMPS Particularly, RF & bias supplies

GMPS regulation

-1.5

-1

-0.5

0

0.5

1

1.5

2

-10 -8 -6 -4 -2 0 2 4 6 8 10

Time (s)

Bias power to 1 Station (kW)

Error in minimum Current


31

GMPS Over a Cycle Train

GMPS regulation

-1.5

-1

-0.5

0

0.5

1

1.5

2

-8.5 -8.3 -8.1 -7.9 -7.7 -7.5 -7.3 -7.1 -6.9 -6.7 -6.5Time (s)

Bias power to 1 Station (kW)

Error in minimum Current

12 12 14 14 1D 1D 1D 1D 1D 1D 1D


32

GMPS Effects• These are all the second $14

w.r.t. a first $14• All display an initial slippage

Starts after ~1 msRate of ~ 7 buckets/msTurns around after notch

• Difference depends on position on trainWorse for NuMI cyclesDepends on the rest of the timelineDifferent signature than other slips

• Hurts PredictionCan’t be fixed by frequent tripplans

• What’s Happening:Beam is bunched into 38 MHZ buckets

• Set externally• No initial Slippage

Phase feedback turns on quickly• Lower magnet current• Beam is pushed toward outside• Slippage begins

Radial feedback turns on after few ms• Gain ramps up• Beam moved to correct position• Slippage lessons quickly


33

What to do with GMPS?• Compensate with cogging:

Cogging “works” with this variation• However, beam is pushed more than we would like

Adequate for moderate intensities

• Fix GMPS:EE Support is looking into feedback

• Feedback should be able to be faster

More stable GMPS might be good all around

• Isolate GMPSMove GMPS to a different feeder than pulsed devicesUnder investigation


34

Half a Notch

• Notcher power supply is old and decrepit

• Notch is marginal on batch trains

• Notch is partial with cogging 5-10% remains in two

buckets Only two buckets wide

• Makes cogging hard We could always go with a

wider notch, but we don’t want to

http://www-bd.fnal.gov/cgi-mach/machlog.pl?nb=numi&action=view&page=-665&button=yes&invert=yes


35

Summary• Proton demand at FNAL has been growing, and will continue to

Current experiments need more protons NuMI is about to start using protons Potential experiments need even more protons

• FNAL has an ongoing process of upgrade and improvement to its proton source Focuses on the Booster and Main Injector accelerators

• Improvement is expected in the Booster

• Main Injector will start high-intensity operations

Improvements have been made with outside collaboration:• Universities, involved in experiments and otherwise

• Other labs (KEK, BNL, LANL, RAL, etc.)

• Until a Proton Driver materializes, improvements in proton intensity will require investments throughout the accelerator complex

Documents

Booster Cogging: Synchronization with the Main Injector for NuMI & Slip-Stacking Robert Zwaska University of Texas at Austin Budker Seminar July 27, 2004